Data Science/Software Engineer

Jayalakshmi Sureshkumar

***** ******* **, *** *, St Louis, MO 63146 ***********.*******@*****.*** 314-***-**** https://www.linkedin.com/in/sureshkumarjayalakshmi https://github.com/jaya81192/AcademicProjects Highlights

Extensive data exploration and machine learning experience using Python, R and Weka in multiple projects.

Well-versed with document indexing and searching in Java using lucene

Advanced knowledge in information retrieval, classification, clustering, reinforcement learning, feature extraction and reduction, genetic algorithm, search algorithms, rule based system

Basic knowledge in natural language processing and an interest to cultivate more knowledge in the same

Excellent SQL knowledge

Education

Indiana University-Bloomington, Bloomington, IN, USA, (GPA 3.83/4.00) Aug 2014-May 2016 Master of Science, Computer Science

Anna University, Chennai, India, (GPA 8.22/10.00) Jul 2010-May 2014 Bachelor of Technology, Information Technology

Skills

Languages: Java, Python, Shell Scripting, C++, C, R, SQL Database: MySQL, Oracle Web: HTML, CSS, Bootstrap, JPA, Spring, Rest Other: Git, Informatica DVO, Weka Packages: pandas, NumPy, scikit-learn, word2vec, matplotlib, lucene, CImg Work Experience

Research Assistant, Indiana University-Bloomington (Java, UMLS Terminology services, MS Excel, Git) Aug 2015-May 2016

Researched and worked on behavioral science research data to find methods that will reduce the uniqueness and granularity within the data set, lowering the risk of re-identification while preserving the semantic meaning of the text

Analyzed concept graph using Java, graph built from concepts and relationships retrieved from UMLS for terms in a field of data describing symptoms

Based on a set scoring, determined amount of aggregation and information loss by aggregation for each term Data Warehouse Intern, World Wide Technology (Informatica DVO, Shell scripting, Oracle 11i, Git, Java) May 2015-Aug 2015

Automated the data validation system for the Data Warehouse with Informatica’s Data Validation Option (DVO)

Assembled audits, configured test creation, deployment and runs; automated testing with schedules

Conducted training sessions for developers and documented detailed curriculum for future reference Associate Instructor, Indiana University-Bloomington (Python, Git) Aug 2014-May 2015

Instructed lab and review sessions for the undergraduate class of Data structures

Graded and assisted in the design of course work Academic Projects

Image Matching, Warping and Stereo (C++, CImg) Feb 2016-Apr 2016

Found matching features between two images using Scale Invariant Feature Transform

Calculated the transformation between two images using RANSAC, and used this to warp one image to the other

Computed depth map with sample foreground and background pixels of an image; created stereo from depth map Object detection with a food image dataset (C++, SVM_multiclass, OverFeat, CImg) Feb 2016-Mar 2016

Detected food using 1250 training images and 250 test images among 25 food category

Extracted features from the image using k mean and using overfeat

Classified using support vector machine algorithm; achieved an accuracy of 44% for k mean and 63% for overfeat features Yelp Dataset Challenge (Java, Lucene) Oct 2015-Dec 2015

Predicted the category of a business in the yelp dataset by indexing and searching review and tip text with lucene

Used pseudo random feedback for improving precision; obtained 86.20% precision and 42.87% recall

Suggested features to be improved for restaurants in two selected areas with significant differences

Clustered businesses into good and bad, and performed feature important through correlation to recommend features

Evaluated using Amazon Mechanical Turk; got average mean square error of .0015 for Montreal and .00076 for Edinburgh Predicting the usefulness of a Review (Python, Word2Vec, Weka) Oct 2015-Dec 2015

Extracted feature from Amazon Review textual data using Word2Vec tool

Clustered using k means into two clusters and labelled based on the 100 manually labelled records; obtained an accuracy of 85% Microsoft Malware Classification Challenge (Java, R, Weka) Mar 2015-May 2015

Performed unigram feature extraction from assembly code using Java and feature reduction using correlation in Weka

Executed Random Forest in R for the classification process; with an accuracy of 98% DivvyApp: A Web Application for Task Distribution (HTML, CSS, Bootstrap, JavaScript, Servlet, MySQL) Aug 2014-Dec 2014

Developed web application using agile model to split household chores among roommates

Worked on the user interface of the application, and on the algorithm and implementation of some of the tasks assigned; the application ultimately helped entering chores into the system, split chores among the group members, kept track of points of each user, carried over points and task from previous week to the current

Contact this candidate